From 97be5890e2a34036a22d2d1e2586c83009ae6064 Mon Sep 17 00:00:00 2001
From: Andrew Butcher <abutcher@redhat.com>
Date: Tue, 12 Jan 2016 16:39:06 -0500
Subject: Validate pacemaker cluster members.

---
 playbooks/common/openshift-master/restart.yml       | 13 +++++++++++++
 playbooks/common/openshift-master/restart_hosts.yml | 15 +++++++++++++--
 2 files changed, 26 insertions(+), 2 deletions(-)

(limited to 'playbooks/common')

diff --git a/playbooks/common/openshift-master/restart.yml b/playbooks/common/openshift-master/restart.yml
index 7603f0d61..fa13a64cb 100644
--- a/playbooks/common/openshift-master/restart.yml
+++ b/playbooks/common/openshift-master/restart.yml
@@ -73,6 +73,7 @@
     register: active_check_output
     when: openshift.master.cluster_method == 'pacemaker'
     failed_when: active_check_output.stdout not in ['active', 'inactive']
+    changed_when: false
   - set_fact:
       is_active: "{{ active_check_output.stdout == 'active' }}"
     when: openshift.master.cluster_method == 'pacemaker'
@@ -98,6 +99,18 @@
     with_items: "{{ groups.oo_masters_to_config | default([]) }}"
     when: (hostvars[item]['current_host'] | default(false)) | bool
 
+- name: Validate pacemaker cluster
+  hosts: oo_active_masters
+  tasks:
+  - name: Retrieve pcs status
+    command: pcs status
+    register: pcs_status_output
+    changed_when: false
+  - fail:
+      msg: >
+        Pacemaker cluster validation failed. One or more nodes are not online.
+    when: not (pcs_status_output.stdout | validate_pcs_cluster(groups.oo_masters_to_config)) | bool
+
 - name: Restart masters
   hosts: oo_masters_to_config:!oo_active_masters:!oo_current_masters
   vars:
diff --git a/playbooks/common/openshift-master/restart_hosts.yml b/playbooks/common/openshift-master/restart_hosts.yml
index 598e1ad63..ff206f5a2 100644
--- a/playbooks/common/openshift-master/restart_hosts.yml
+++ b/playbooks/common/openshift-master/restart_hosts.yml
@@ -16,8 +16,6 @@
       delay=10
       port="{{ openshift.master.api_port }}"
   when: openshift.master.cluster_method != 'pacemaker'
-# When cluster_method is pacemaker we can only ensure that the host
-# restarted successfully.
 - name: Wait for master to start
   become: no
   local_action:
@@ -25,4 +23,17 @@
       host="{{ inventory_hostname }}"
       state=started
       delay=10
+      port=22
   when: openshift.master.cluster_method == 'pacemaker'
+- name: Wait for master to become available
+  command: pcs status
+  register: pcs_status_output
+  until: pcs_status_output.stdout | validate_pcs_cluster([inventory_hostname]) | bool
+  retries: 15
+  delay: 2
+  changed_when: false
+  when: openshift.master.cluster_method == 'pacemaker'
+- fail:
+    msg: >
+      Pacemaker cluster validation failed {{ inventory hostname }} is not online.
+  when: openshift.master.cluster_method == 'pacemaker' and not (pcs_status_output.stdout | validate_pcs_cluster([inventory_hostname])) | bool
-- 
cgit v1.2.3